Office of Inspector General

DEPARTMENT OF HOMELAND SECURITY Office of Inspector General USCIS Approval of H-1B Petitions Exceeded 65,000 Cap in Fiscal Year 2005 Office of Inspections and Special Reviews OIG-05-49 September 2005

Office of Inspector General U.S. Department of Homeland Security Washington, DC 20528 Preface The Department of Homeland Security (DHS) Office of Inspector General (OIG) was established by the Homeland Security Act of 2002 (Public Law 107-296) by amendment to the Inspector General Act of 1978. This is one of a series of audit, inspection, and special reports prepared by our office as part of our DHS oversight responsibility to promote economy, effectiveness, and efficiency within the department. This report assesses the processing of H-1B "temporary worker" nonimmigrant visa petitions by DHS in fiscal year 2005 and analyzes the deviation between the statutory ceiling and the actual results. It is based on interviews with employees and officials of relevant agencies and institutions, direct observations of the petition process, and a review of applicable documents. The recommendations herein have been developed to the best knowledge available to our office, and have been discussed in draft with those responsible for implementation. It is our hope that this report will result in more effective, efficient, and economical operations. We express our appreciation to all of those who contributed to the preparation of this report. Richard L. Skinner Inspector General

Contents Executive Summary... 1 Background... 1 Results of Inspection... 3 The traditional CIS business process guarantees imprecision... 3 The organization of DHS handicaps counting efforts... 5 A complex adjudication process makes the cap count fluctuate... 5 A complex counting process makes the cap a moving target... 9 An unexpected influx of petitions swamped the cap counting process... 15 Ceasing acceptance of new petitions took too long... 17 Recent CIS initiatives are intended to prevent a recurrence... 19 Conclusion and Recommendations... 20 Management Comments and OIG Analysis... 22 Appendices Appendix A: Purpose, Scope, and Methodology... 24 Appendix B: Section 214(g) of the Immigration and Nationality Act... 25 Appendix C: Management Response to Draft Report... 28 Appendix D: Major Contributors to This Report... 33 Appendix E: Report Distribution... 34 Abbreviations ACWIA American Competitiveness and Workforce Improvement Act AC21 American Competitiveness in the 21st Century CIS U.S. Citizenship and Immigration Services CLAIMS Computer Linked Application Information Management System DHS Department of Homeland Security DO Office of Domestic Operations, CIS DOS Department of State KCC Kentucky Consular Center LAN Local Area Network LCA Labor Condition Application OIG Office of Inspector General OIS Office of Immigration Statistics RFE Request for Evidence

OIG Department of Homeland Security Office of Inspector General Executive Summary In March 2005, the Office of Inspector General received a letter from Senator Charles Grassley, Chairman of the Senate Finance Committee, and Representative John Hostettler, Chairman of the Subcommittee on Immigration, Border Security and Claims of the House Judiciary Committee. The chairmen requested that we investigate the actions taken by officials of U.S. Citizenship and Immigration Services (CIS) regarding the bureau's provision of H-1B non-immigrant status to more aliens in fiscal year (FY) 2005 than was statutorily authorized. In particular, we were asked to investigate how the over-issuance of H-1B visas occurred and whether it was done in deliberate violation of federal law. In addition, we were asked to make determinations as to what actually transpired and how to prevent an over-issuance from occurring in the future. In response to their request, we reviewed the H-1B petition approval process and the events that led to this overissuance in FY 2005. CIS officials at all levels in Washington, DC and at the service centers were aware of and attempted to comply with the statutory limit on the number of persons granted H-1B status. However, CIS had neither the technology nor an operational methodology to ensure compliance with the precise statutory ceiling. Faced with the certainty of issuing either too few or too many approvals, it had been CIS' explicit practice to avoid approving too few. The CIS "business process," of taking all petitions submitted before an announced cut-off date, guarantees that an inexact number of petitions will be approved. The structure of DHS handicaps counting efforts; a complex adjudication process makes the count fluctuate; a complex counting process makes the cap a moving target; and, an unexpected influx of petitions in mid-september 2004 swamped the cap counting process. Several recent CIS initiatives are designed to prevent a recurrence. We cannot evaluate policies meaningfully that have not yet been tested. However, we believe that they might not be sufficient to accomplish the precision that Congress now requires, and offer two recommendations to improve the methods for processing H-1B petitions. Background The H-1B program is one of several "temporary worker" nonimmigrant visa programs. It allows U.S. based companies to employ foreign individuals in the Page 1

United States for specialty occupations on a temporary basis. There are three subcategories: H-1B1 positions - for workers from Chile and Singapore - require that the alien have at least a bachelor s degree or the equivalent in a specialty field; H-1B2 positions - for workers who will perform exceptional services for a project administered by the US Department of Defense (DOD); and, H-1B3 positions - for fashion models of national or international acclaim and recognition. Other temporary worker visa programs are designed to accommodate trainees, intra-company transferees; aliens of extraordinary ability in arts, science, education, business or athletics; internationally recognized athletes and entertainers; and, aliens coming temporarily to participate in an international cultural exchange program. Both the entertainment industry and professional sports employ many temporary workers. Section 214(g)(1)(A) of the Immigration and Nationality Act sets a limit of 65,000 H-1B approvals in FY 2005. The same section of the law defines certain types of H-1B petitions as exceptions to this "cap." Therefore, CIS must determine as it adjudicates each petition whether it is a "cap case" or a "non-cap case." There is no limit on the number of cap-exempt petitions that can be approved. The law requires, as well, that H-1B cases be approved in the order in which they were filed. Section 214(g) is complex; the complete text is attached at Appendix B. In FY 2005, CIS exceeded the limit and approved more petitions than the law stipulated. 1 Once adjustments were made to the total number involving visa denials and reconsidered exemptions, and after unused numbers set aside earlier to comply with free trade agreements were factored in, the total number of approved FY 2005 "cap cases" turned out to be approximately 71,740 -- significantly over the limit of 65,000. 2 CIS adjudicators examine many factors before approving an H-1B petition. Both the position that is going to be filled and the worker who will fill it must meet many criteria, all of which are described in the service center operations manual. Petitions that are complete and clearly meet the standards can be quickly approved. Other petitions require correspondence between the service center and the petitioner to resolve unclear or incomplete submissions. 1 CIS "White Paper" dated April 12, 2005. 2 CIS-OIG e-mail June 27, 2005. Page 2

H-1B petitions, when approved, are used by some beneficiaries to apply for an H- 1B visa at an embassy abroad. Other beneficiaries, already lawfully in the United States in some other status (such as F-1 student) need to change status with CIS. Canadian beneficiaries are reviewed for admission when they arrive at the border, because Canadians are exempt from the visa requirement. The cap does not limit H-1B visas, as generally is believed, but limits persons granted H-1B status, which complicates the counting process. Results of Inspection CIS approved approximately 72,000 H-1B cap cases for FY 2005, exceeding the statutory cap of 65,000. There is no indication that this was done in deliberate violation of federal law. A number of factors led to this outcome. The traditional CIS business process guarantees imprecision CIS cannot exactly hit any specific cap number with its current business process. CIS management (and formerly INS) has understood that the final number of cap cases approved would be higher or lower than the exact mandate. CIS makes many efforts to minimize this deviation, but believes that the public is best served by simplicity and transparency in the petition process. CIS cuts off further receipt of H-1B petitions when it estimates the number of approved petitions is approaching the mandated limit. CIS makes efforts to monitor its H-1B workload, to count the incoming cap cases, and to estimate the likely outcome of the pending work already received but not yet finally adjudicated. These efforts allow CIS to announce a cut-off date after which it will no longer accept petitions so as to end the year close to the cap. All petitions received before the cut-off dates are adjudicated to completion without regard to the cap. CIS managers told us that American employers are entitled to be confident that petitions that meet the legal requirements and are filed in a timely manner will be approved. We interviewed an official who was involved with the H-1B cap issue in the late '90s at the former INS. She indicated that a perceived failure at that time to "hit the cap" was brought to the attention of the commissioner, and that efforts were undertaken to improve management's ability to approve the maximum number of allowed petitions. Among other initiatives, the form petitioners complete and submit, the Form I-129, was redesigned to better capture information relating to whether a case was exempt from the cap. When INS failed to hit the cap again the following year, outside consultants were engaged to take another look at the process. Alternative petition processing systems have been discussed but dismissed as inferior to the current process. Page 3

CIS has considered and rejected a business process that would involve opening and closing the application window repeatedly during the 18-month processing period that can be used each fiscal year. 3 During each closure CIS could adjudicate all pending cases, and evaluate how many more cap cases could be processed. Also, it considered and rejected a model in which it would accept and process cases until the cap is reached, and then mail back the unprocessed overage along with the paid fees. Our visit to the mail room at the California service center, where petitions of all kinds arrive by the truckload several times a day, gave us an appreciation of how difficult such an alternative would be. Also, it might expose CIS managers to significant internal control issues with respect to a potentially large number of fees. CIS told us it seeks a process that will avoid cutting off further applications too soon before the statutory limit is reached. Businesses in the past have brought legal actions against CIS when they believed CIS over-counted and, therefore, failed to approve the full statutory number of cases. 4 CIS today continues to believe that any shortfall in petition approvals that would result from an over-counting of cap cases would be a disservice to the American business community. Ironically, in mid-september, 2004 -- at almost the exact moment that CIS was inadvertently going over the FY 2005 cap -- a manager sent an e-mail discussing with her colleagues and subordinates the possibility that there had been a shortfall in FY 2004. 5 She indicated that "there are concerns that perhaps some part of the 'formula' [for predicting eventual approvals] is not correct in that perhaps we did not capture some group of cases that are falling out of the cap usage classification." After mentioning several possible explanations for the inaccuracy of the FY 2004 count, she said: " [our superiors] want to make sure that we revisit this formula so we can get closer to the target number before we shut off the H-1B cases for FY05." A fuller description of the adjudication process has been provided in a later section. Such a description is necessary to fully explain the complexities that result in cap cases reclassified as non-cap - or vice versa - during processing. Approved cases can later be revoked, freeing an additional number. Denied petitions can be appealed successfully to a higher level in CIS, consuming a number. Beneficiaries believed to be different persons (and counted twice) can in fact be one person with multiple offers from potential employers. Beneficiaries abroad are sometimes denied their visa - freeing another cap number if they were a cap case and if CIS is informed of the denial. For these and other reasons, the number of already approved cap cases would fluctuate even if CIS temporarily 3 Petitions can be submitted up to six months before the beginning of each fiscal year, i.e., April 1. 4 See, for example, Law Office of Azita Mojarad, et al v. Eduardo Aguirre, Case number 1:05CV00038, U.S. District Court for the District of Columbia. 5 September 15, 2004 e-mail titled "H-1B Cap Counts". Page 4

ceased adjudicating pending cases. With the number of petitions already approved and subject to the cap in constant fluctuation, it is almost impossible for CIS to approve any specific number of cap cases. The organization of DHS handicaps counting efforts Before DHS was created, the Immigration and Naturalization Service (INS) Office of Immigration Statistics (OIS) played an important role in managing the cap count by providing timely H-1B workload numbers to INS managers. When DHS was formed OIS was not placed within any of the three new fragments of the former INS: Customs and Border Protection, Immigration and Customs Enforcement, or Citizenship and Immigration Services. Instead it was made a component of the DHS Office of Management, one of the five major directorates that report directly to the Secretary. This left CIS responsible to administer a process it could not monitor with sufficient precision on its own. During the first two years of DHS existence OIS continued to provide counting assistance to CIS managers. We reviewed correspondence between CIS and OIS and found that the offices cooperated well as they tried to keep an up-to-date count of H-1B activity. Nevertheless, as one interviewee described the situation to us, CIS had the responsibility to manage numbers but no ability to track them, while OIS could track the numbers but did not have responsibility to manage the H-1B program. Further complicating the relationship, and slowing down the count, was the need to utilize an outside contractor to process the data and generate the count estimates. The inadequacies of the Computer Linked Application Information Management System (CLAIMS) used by CIS will be discussed below, but we found that CIS managers required both the assistance of OIS and the paid services of a contractor to generate the numbers they needed to manage the cap. A significant improvement was made in December 2004. An expert in H-1B statistics was transferred from OIS to CIS. CIS expects in the future to be able to manage the cap count without depending upon OIS for routine cap counting assistance. A complex adjudication process makes the cap count fluctuate H-1B petitions need to be examined, evaluated, and adjudicated. Errors, inconsistencies, and omissions from the file need to be resolved. Denials are subject to administrative appeal. Approvals are subject to later revocation. CIS promises an expedited decision to those who pay a surcharge. Overall, the business processes are stunningly unwieldy. Page 5

Petitioning employers file Form I-129, Petition for Nonimmigrant Worker, with one of four CIS service centers at Laguna Niguel, California; St. Albans, Vermont; Dallas, Texas; or Lincoln, Nebraska. Petitions for workers who seek specialty occupations (the bulk of H-1B petitions) must include supporting material that proves eligibility for H-1B status. This evidence includes a Labor Condition Application (LCA) from the Department of Labor and a prevailing wage determination received from either the state workforce agency that has jurisdiction over the geographic area of intended employment or from a survey conducted by an independent authoritative source. Evidence also is required to prove the beneficiary has at least a U.S. baccalaureate, a comparable foreign degree, or education or experience that is equivalent. Required processing fees, state employment licenses, and a contract of employment also must be supplied. CIS service centers date stamp each petition, deposit the submitted fees, and create a paper file. A corresponding computer record with information about the petition is established in the CLAIMS Local Area Network (LAN). 6 A unique receipt number is generated for each application. The paper files are then sent to the file control unit where they are sorted into cap and non-cap cases. This sorting is performed to allow for continued processing of non-cap cases at the service centers once the annual cap has been reached. Files are then sent to the adjudicators based on local workload requirements. CIS adjudicators review the files (both the paper and electronic versions) and determine whether there is sufficient evidence to determine H-1B status eligibility. 6 The CLAIMS LAN provides users, including supervisors, with basic automated support for: a) adjudicating cases, b) processing various notices, c) producing supporting documentation (e.g., signature cards) for files supported by the system, and d) administrative functions related to the use of the system in support of the H-1B process. The CLAIMS LAN is a hybrid application, with extensive functionality distributed between the Service Centers and the CLAIMS Mainframe in Dallas, Texas. The LAN-based version of CLAIMS implemented at each Service Center is tightly coupled with a mainframe-based version of the same application. The local version of CLAIMS supports end users at each of the Service Centers through four separate systems which are: a) data entry subsystem, b) adjudication subsystem, c) processing support subsystem, and d) automatic data processing system. Page 6

Incoming Mail at the California Service Center Adjudicators will reject a petition if it is incomplete, or lacks appropriate fees or required supplemental material. In such cases, CIS sends a Request for Evidence (RFE) to the petitioner, asking for any additional information required to support the petition. The petitioner must respond within a set period of time or the petition will be considered abandoned and will be denied. If the petitioning employer supplies the requested evidence, the adjudication resumes. Cap cases are required by law to be approved in the order they were filed (see Appendix B - 214(g)(3)), so if the incomplete case is or might be a cap case, a cap number must be assigned to the petition while the RFE is outstanding in case the employer does provide the evidence and the petition is approvable. A petition can be denied if the adjudicator decides the worker is not eligible for H-1B status. A denial does not necessarily end the petitioning process. The company can file an appeal within thirty days. Appeals and motions are handled by the CIS Administrative Appeals Office, located in Washington DC. As in the event of an RFE, if a case is on appeal a cap number must be held aside so that if the appeal is successful a number will remain available. 7 7 The Administrative Appeals Office estimated for us that appeals succeed petitions at first denied are later approved approximately 15 % of the time, and that several thousand appeals are submitted each year. Page 7

If an adjudicator finds that a worker is eligible, the petition is approved and the petitioning company is sent Form I-797, Notification of Approval. If the beneficiary is in the United States, the alien is adjusted to H-1B status. If the worker is outside of the United States, an I-797 is also sent to the Department of State Kentucky Consular Center (KCC), which notifies the appropriate U.S. embassy abroad. The worker (unless a Canadian citizen) must make an appointment with the embassy and apply for an H-1B visa. Visa law is complex, and it is not uncommon for a beneficiary approved by CIS to be found ineligible to receive a visa by the Department of State. In that event, if the petition was subject to the cap, the cap number can be reassigned for the benefit of another petition. 1) Company sends petition to designated Service Center via mail or internet USCIS National Service Centers Petitions can be: Rejected 2) Adjudication takes place at one of four Service Centers, and typically takes 30 days Incomplete petitions are returned with a Request for Evidence (RFE). Reply required within 81 days Company Planning to Employ Worker Denied If the petition is denied an appeal can be made within 30 days 4) For workers in the U.S., worker is adjusted to H-1B status without a visa and may begin employment Beneficiaries outside the U.S. make appointment with visa officer at consulate 3) If approved, company is notified with Form I-797. Worker Local US Embassy Approved 5) Visa officer determines if worker is eligible for H-1B visa. If approved worker uses visa to enter the U.S. and begin employment 4) For workers abroad, approval file is sent through the DOS KCC to appropriate embassy DOS Kentucky Consular Center Figure 1 The H-1B Application Process Page 8

The process for adjudicating an H-1B petition can take a CIS service center up to sixty days, though most petitions are processed within thirty days. RFEs for petitioners can slow the process considerably. For an additional fee, petitioning companies can use an expedited service that guarantees adjudication within two weeks. With this expedited service, some approvals are made within hours of the CIS service center receiving a petition. A complex counting process makes the cap a moving target In addition to the challenges already discussed, hitting the cap target is made even more difficult by exemptions in the law and by the procedures CIS applies to keep a running total of cap cases approved to date. Many Legal Variables Affect the Cap In 1990, Congress amended the Immigration and Nationality Act to impose a cap on the numerical limit of aliens who may be granted H-1B status at 65,000 per fiscal year. In 1998, Congress passed the American Competitiveness and Workforce Improvement Act (ACWIA), which increased the cap to 115,000 for FY 1999 and FY 2000, and to 107,500 for FY 2001. In 2000, Congress passed the American Competitiveness in the 21st Century Act (AC21), which increased the cap to 195,000 for FYs 2001, 2002 and 2003. In FY 2004, the cap returned to 65,000 per year. Fiscal years begin in October of the previous calendar year, that is, October 1, 2004, through September 31, 2005, is FY 2005. But because petitions may be filed six months before a potential worker s intended start date, a company may file a FY 2005 petition as early as April 1, 2004, for a worker who will begin work on October 1, 2004. The latest a company can file is dependent on whether the congressionally-mandated cap has been reached. If the cap has not been reached for a given fiscal year by the time petitions are accepted for the next fiscal year (April 1), a worker can be applied to either fiscal year. CIS must make several other adjustments to the cap count during the petition process. For example, multiple petitions may have been submitted for one beneficiary, such as a college senior who is offered employment by more than one company. Each petition will appear to be a separate cap case. If CIS fails to adjust its count, it will incorrectly believe it is using more cap numbers than in fact it is. CLAIMS does not automatically detect multiple petitions for one beneficiary, so the data must be reviewed periodically through a special process. Another exemption from the cap is for any beneficiary who has been counted against the cap within the last six years (see Appendix B, 214(g)(7)). Eligibility for this exemption is based upon the beneficiary's previous immigration history, which often is not known to the prospective employer completing the petition Page 9

form. Therefore, in cases where this exemption applies, it is often not apparent to the adjudicator from the information in the petition. Neither, officials told us, would the CLAIMS LAN indicate to an adjudicator that the same-named beneficiary was also the subject of an earlier petition processed at any of the other three service centers. Determining whether or not 214(g)(7) applies can be particularly difficult for CIS. In addition, the several other statutory exemptions (see Appendix B, 214(g)(5)) from the cap need to be excluded from the count. These benefit employers who are institutions of higher learning and any affiliated nonprofit entities, and other nonprofit or government research organizations. The Chile and Singapore Free Trade Agreements of 2003 reserve 6,800 of the 65,000 H-1B slots for workers from Chile and Singapore (1,400 for Chile and 5,400 for Singapore.) Unused Chile or Singapore numbers the demand for which so far has been small - are restored to the larger pool of 58,200. This is difficult to implement, however, because a number is not definitively unused until the last day of the fiscal year. The law therefore provides that the unused Free Trade Agreement H-1B numbers have a special 45-day window in the beginning of the following fiscal year to be used against the cap of the just-completed fiscal year. All of the other 58,200 cap numbers belonging to a fiscal year, however, can actually be used up before the fiscal year even begins during the six-month lead-in period described earlier. Previously approved cap cases that subsequently are revoked need to be subtracted from the total, as do visa refusals abroad, so that the unused number can be given to another petition. To give an example, in FY 2004 slightly over 500 H-1B visa applications abroad were denied. When to cut off further receipt of petitions? The CIS Office of the Associate Director, Domestic Operations (DO) is responsible for enforcement of the annual cap. Cap counting activity is managed by Service Center Operations (SCOPS), a DO component. Within SCOPS the deputy director is responsible for managing the H-1B cap. This responsibility includes "regular monitoring of the volumes of approved and pending petitions subject to the cap and the manner in which USCIS determines the number of pending petitions that will likely result in additional approvals subject to the cap." 8 With the assistance of OIS, a special process has been developed to monitor H-1B volume. A contractor extracts data from CLAIMS, makes some historically based adjustments to allow for probability of various outcomes, and analyzes the results. 8 Source: declaration submitted by CIS in the matter of Law Office of Azita Mojarad, et al v. Eduardo Aguirre ( see Footnote 5). Page 10

Each repetition of this process is called a sweep, and requires about two days. Each sweep requires retrieval of data from all four service centers. In the early part of the year a sweep might be performed once every two weeks, but, as the cap is approached, the frequency increases. DO uses sweep data, along with previous years trends, to determine a termination date for accepting new petitions. While the exact formulas are complicated, the process basically works as follows: a known number of cap cases is approved. To this number is added an estimated fraction of the pending work at the service centers the cases received but not yet adjudicated. An addition is made for an estimated percentage of the cases denied but still on appeal. A subtraction is made for an estimated percentage of cases that will be denied visas. Another adjustment is made for claims to be cap-exempt that later will be determined to be unfounded. Numerous other predictions are factored in, and the result is an estimate of the number of cap cases likely to be approved from all the petitions received as of two days ago. With this number, and in light of the rate at which new petitions arrive each day, management can estimate when it will suspend processing because of the cap. Once this date is determined, a public announcement to that effect is issued. One CIS official recently explained the challenge this way in an e-mail memo 9 : Petitions Pending Adjudication -- Unadjudicated or pending petitions present another counting challenge. At any one moment thousands, usually tens of thousands of cap-eligible petitions may be awaiting adjudication. The management of the cap requires predicting the adjudication outcome of these petitions as well as any additional information on exemptions. Some of these pending petitions will be denied and/or become exempt from the cap. However, a decision on when sufficient petitions have been received to satisfy the cap will be made weeks before the cap is actually reached.... How close in theory the final cap count is to the statutory limit depends to a large extent on the accuracy of the statistical model used to predict the outcome of the adjudication process of these pending petitions. The statistical model employs assumptions based on the experience of petitions already adjudicated in the same fiscal year. The past is not necessarily an accurate predictor of the future, certainly not if an exact count is required. In another e-mail, a week later, the same official shared a spreadsheet to illustrate the effect that different assumptions about the outcome of pending cases would have on estimating a correct cut-off date. He said this about the data: Attached is a file showing the daily cap count for the 10-day period beginning September 13. The table illustrates the difficulty of managing the 9 3/31/2005 SCOPS e-mail titled "H-1B Cap Counting Issues and Complexities" Page 11

cap when such a large number of cap-eligible petitions (between 23,600 and 32,200) are pending adjudication. When the cap was reached depends on what is assumed about the adjudication outcome of these petitions. The FY 2005 cap was apparently reached between September 16th and 22nd. I will endeavor to figure exactly which day it was reached. That will also reveal how (un)reliable the assumptions were. The size of the pending cap is a function of cap-eligible petition inflows and adjudication time. The greater the inflows, and the slower the adjudication process, the faster pending grows. The surge of petitions in September made predicting the date the cap would be reached that much more difficult since adjudications did not match the pace of incoming petitions. The spreadsheet illustrates that the cap will be projected to be reached on different dates depending on assumptions made about the large body of pending cases. He shows how three different real past outcomes - any one of which might be reasonably considered to be a predictor of future approval rates - yield three different cut-off dates. The three possible projections he uses that 67, 81, or 90 percent of pending cases which will eventually be approved and found to count against the cap are based on specific measures of past approvals. FY 2005 Cap count Cap Cap count Approved Cap Pending Cap Potential Cap Estimate #1 Cap Estimate #2 Cap Estimate #3 as of (A) (B) (A)+(B) (A)+(.67*(B)) (A)+(.81*(B)) (A)+(.90*(B)) September 13 38,498 23,612 62,110 54,318 57,624 59,749 September 14 39,185 24,659 63,844 55,707 59,159 61,378 September 15 39,886 25,744 65,630 57,134 60,739 63,056 September 16 40,670 27,042 67,712 58,788 62,574 65,008 September 17 41,370 28,666 70,036 60,576 64,589 67,169 September 18 41,598 28,727 70,325 60,845 64,867 67,452 September 19 41,669 28,763 70,432 60,940 64,967 67,556 September 20 42,337 30,813 73,150 62,982 67,296 70,069 September 21 42,543 30,817 73,360 63,190 67,505 70,278 September 22 44,429 32,187 76,616 65,994 70,500 73,397 Figure 2 CIS E-mail Shows Criticality of Assumptions What information did CIS have at the time? We have reviewed the many e-mails in which CIS officials reported the results of each of the sweeps described above. In each status report the sender provided three numbers: the number of cap petitions approved to date; an estimated number of cap cases that would likely be approved from the work already received but not adjudicated; and the total of those two to estimate how close CIS was to reaching the cap. Page 12

In the beginning of the fiscal year sweeps were performed less frequently, because there was little cause to believe the cut-off date was approaching. As the sweep data showed higher and higher numbers, the frequency of the sweeps increased to the point that they were executed every week. Table 1 shows the results of the sweeps. Table 1: Cap-eligible and Cap approved petitions for FY 2005 10 As of 11 Cap approvals 12 (A) Cap-eligible pending adjudication (B) Potential Cap Total A + B 06-09-04 7,500 11,500 19,000 06-16-04 8,700 13,200 21,900 06-23-04 10,000 14,400 24,400 06-30-04 11,300 15,800 27,100 07-21-04 15,600 18,700 34,300 07-28-04 18,100 18,800 36,900 08-04-04 21,000 19,900 40,000 08-11-04 24,100 19,100 43,200 08-18-04 27,300 18,600 45,900 08-25-04 30,100 19,000 49,100 09-01-04 32,900 20,100 53,000 09-17-04 40,000 27,400 67,400 09-24-04 44,600 31,600 76,200 09-29-04 48,000 32,700 80,700 10-01-04 49,900 33,000 83,200 10-06-04 52,600 33,300 85,900 10-14-04 56,700 29,800 86,500 10-21-04 50,700 13 26,800 77,500 10-28-04 53,300 24,100 77,400 11-04-04 55,200 22,200 77,400 12-02-04 60,800 16,000 76,800 12-09-04 62,000 14,700 76,700 01-07-05 64,600 11,600 76,000 01-20-05 66,800 8,900 75,700 01-27-05 68,000 7,400 75,400 02-03-05 69,000 6,100 75,100 The second column indicates the number of cap cases that have been approved as of the date of the data sweep. The third column is the calculated estimate of the number of approvable cap cases that will eventually come out of the not-yetadjudicated cases then pending at the service centers. The fourth column indicates the projected number of cap cases that will be approved once the current pending work is eventually adjudicated. 10 Source: CIS. 11 Refers to date extract file was created. 12 The cutoff date for cap receipts was February 17, 2004. 13 Adjusted for 8,700 individuals rolled back to FY 2004. Page 13

How reliable are CIS estimates? The bottom portion of the fourth column of Table 1 provides a clear illustration of how difficult it is to estimate the outcome of cases not yet adjudicated, pending an RFE, or on appeal. The data sweep of October 14 showed that CIS was likely to approve 86,500 cap cases. After the receipt of additional cases ceased, adjudicators worked through the backlog, evidence previously requested was received, and appeals were decided. In the weeks that followed, the number of estimated total cases dropped every week but one. The previous estimates about the outcome of these cases proved to be quite inaccurate. By January 7, almost three months after the projected 86,500 number had been set, CIS had not achieved 65,000 approvals. In the final analysis, as we now know, the total of approved cap cases reached just slightly over 72,000. FY 2005 was not the first time that the cap was exceeded. The cap was also surpassed in FYs 1998, 1999, and 2000: In FY 1998, INS approved roughly 19,000 more petitions than the 65,000 allowed. Because the cap was met late in the fiscal year, the excess petitions were counted against the subsequent fiscal year ceiling. In FY 1999, the cap was raised to 115,000 but was again exceeded partly because of the rolled-over petitions from FY 1998. After accepting too many petitions over two years in a row, INS contracted for a review of the cap counting process by the consulting firm KPMG. In FY 2000, the 115,000 cap was met, and the remainder was counted toward the next fiscal year. The cap was raised to 195,000 for FYs 2001, 2002 and 2003. It was not exceeded. In FY 2004, the cap reverted to 65,000. Receipt of petitions was cut-off on February 17, 2004 in anticipation of hitting the cap. Once all the FY 2004 petitions were processed, it was determined that approximately 63,000 cap cases were approved - 2,000 fewer than the statutory limit. In October 2004, once it was clear that the FY 2005 cap had been exceeded, 2,000 approved cases received between April 1 and September 30, 2004 were "rolled back" and deemed to count towards the FY 2004 cap. Table 2 shows the cap, the number of cap cases actually approved and the date processing of further cases was cut-off from FY 1998 through FY 2005. Page 14

Table 2: Cap, Cap Approvals and the Cut-Off Date FY Cap Cap Approvals Cutoff Date 1998 65,000 84,000* None 1999 115,000 137,000 April 16, 1999 2000 115,000 115,000** March 21, 2000 2001 195,000 164,000 None 2002 195,000 79,000 None 2003 195,000 78,000 None 2004 65,000 65,000*** February 17, 2004 2005 65,000 72,000 October 1, 2004 * 19,000 were rolled over to FY 1999. ** Some excess petitions were counted towards the FY 2001 cap. *** 63,000 cases were approved of those received before the February 17, 2004 cut-off. Later, 2,000 cases received between April 1 and September 30, submitted for FY 2005, were rolled back and allocated against FY 2004. An unexpected influx of petitions swamped the cap counting process The CIS Press Office provides the public with periodic updates on changes to immigration law, procedures, deadlines and various immigration statistics. Through such announcements, companies are informed when CIS service centers will stop accepting new H-1B cap petitions. On September 2, 2004, a press release announced that, as of August 18, 2004, CIS had received 45,900 H-1B petitions that would count against the congressionallymandated cap of 65,000 for fiscal year 2005. The rate petitions arrived at CIS mailrooms accelerated markedly immediately after the press release, as is shown in Figure 3 below 14. 14 Although H-1B eligibility differs from that for an L-1 intra-company transferee, an examination of CIS data we performed indicated that 1,975 L-1 beneficiaries had H-1B cases pending in FY 2005, too. Beneficiaries soliciting multiple petitions may have increased the H-1B workload. Page 15

Receipts for Initial H-1B Status 10,000 September 20 9,000 September 13 Number of ReceiptsX 8,000 7,000 6,000 5,000 4,000 3,000 2,000 June 7 July 5 Septem ber 27 Septem ber 6 August 30 August 23 August 16 Weekly receipts 1,000 0 April 1 April 1 April 26 May 24 June 21 July 19 August 16 September 13 Week beginning in 2004 Figure 3 Number of Petitions Received at Service Centers Each Week (Source: CIS) One senior manager we interviewed stated that CIS was surprised by this pattern. Much publicity had surrounded the H-1B cap in the press, and there had been considerable debate on Capitol Hill about the cap levels that best served U.S. economic interests. He said CIS expected a rush of H-1B petitions in April and May. Not only did that not occur, but as the summer season continued there was only a slight rise in the rate at which petitions arrived at service centers. As late as mid-august, he continued, it appeared that the cap might not be met until well into the following calendar year. While the sudden rise in the last week of August and the first week in September can now be viewed as the precursor of a massive surge, it was not recognized as such at the time. On September 17, 2004, the report on the cap count from the CIS service centers projected enough adjudicated and pending petitions to justify a halting of the new petitions. FY 2005 is unique in one respect - in previous years the cap was reached late in the fiscal year. Once April 1 arrives, six months before the October 1 start of a new fiscal year, the petition window opens for the coming fiscal year. Petitions Page 16

received after April 1 can be allocated in two ways: to the current fiscal year if the cap has not yet been reached, or to the following fiscal year if it has. Cases over the cap in FY 2005 could not be rolled over to the next fiscal year because the cap was reached too early six months before the FY 2006 processing began on April 1, 2005. Ceasing acceptance of new petitions took too long CIS could have closed the window sooner than it did if procedures to do so had been approved in advance. In the event, as soon as it became apparent that receipt of petitions had to be suspended, it was necessary to draft a notice to that effect for the Federal Register, and a press release, and circulate both for clearance and approval. This process was delayed by the need to manage two other related issues at the same time: student visa extensions and Chile-Singapore Free Trade Agreement visa procedures. We have analyzed most of the e-mail discussions that took place within CIS as the number of approved H-1B petitions approached, and then exceeded, 65,000. This analysis shows that CIS managers were poorly served by the inability of their data systems to produce real-time numbers, by their own incorrect projections, and by perceived lack of clarity in the Chile-Singapore statutory language. On August 25, 2004, demonstrating all three of these issues at once but working with week-old data, one senior manager wrote to senior colleagues that: "it is not inconceivable that we could exhaust all FY 2005 numbers before fiscal year 05 officially begins. It is more likely that we will need to keep filings open until November 14.... The cap for FY 2005 is somewhat complicated.... Those restrictions for the use of the unused [Chile-Singapore] numbers include a requirement that the numbers must be used within 45 days of the beginning of the fiscal year (November 14, 2004), and a requirement that the numbers can only be used for petitions filed in the fiscal year that they are used (FY 2005 in this case). I have not sought legal review of that language to ascertain whether there is any flexibility on those provisions, but am including [Office of Chief Counsel] in this transmission to seek review and advice". On August 27, 2004, the Office of Immigration Statistics issued a routine cap count e-mail reporting that the sweep two days earlier indicated 30,100 cases approved and 19,000 pending, for a potential total of 49,100. It indicated that the analyst would be out of the office for the next two weeks, too. The first e-mail that circulated a draft Federal Register Notice concerning a cessation of acceptance of H-1B petitions was dated September 10, 2004. On September 15, 2004, the OIS analyst circulated the results of the September 1st sweep. This data was two weeks old. It showed approvals at 32,900 and Page 17

pending cases at 21,100 for a total of 53,000. Also on September 15, 2004, it was reported that the September 8th sweep attempt had failed due to a large amount of corrupted data. On September 20, 2004, the OIS analyst circulated the results of the September 17th sweep indicating that the total had exceeded the cap with 40,000 petitions approved and 27,400 pending. He informed the recipients that there had been a dramatic increase in H-1B receipts in the first two weeks of September. Again on September 20, 2004, senior CIS managers met and decided that adjustments to the count should be made to compensate for 2,000 FY 2004 numbers that had been unused as a result of closing the FY 2004 window a little too soon in February 2004. Also, an adjustment would be made to reuse the 5,700 Chile/Singapore numbers that had been set aside but not used. These two decisions effectively "rolled" 7,700 FY 2005 cases forward into FY 2004. On September 27, 2004, the OIS analyst circulated the results of the September 24th sweep. It showed approvals at 44,600 and pending cases at 31,600 for a total of 76,200. Another e-mail concerning the draft Federal Register Notice was sent on September 28, 2004. The author noted that in light of the latest approval numbers " we probably need to get clearance on the H-1B cap notice for FY 2005 in short order. Please review ASAP and fax signed concurrence sheets to my attention." The draft is nine pages in length and discusses, in addition to the congressional cap, two related issues: the extension of status of students and exchange visitors for whom H-1B petitions had been filed, and the recycling of the unused Chile/Singapore numbers previously set aside. On September 29, 2004, a senior manager in Service Center Operations sent an e- mail suggesting that CIS cease accepting H-1B petitions subject to the cap on October 1, 2004. The e-mail discussed the latest view of the proper treatment of pending cap cases. CIS knew after each sweep how many H-1B cap cases it had already approved, and it knew how many petitions had been received but not yet finally adjudicated. The number of pending cases that would eventually be approved was estimated in order to assess the total number of cap cases that would be approved. In late September the analyst sought to improve his statistical model. He knew that prior year estimations had not been very precise, so he added new variables to the mix. Based on this finer analysis, he advised CIS managers that the September 24th sweep data should be reinterpreted. Instead of assuming that 31,600 of the pending cases would be approved, his data now suggested that 21,800 would be. If correct, this in turn meant that CIS was likely to approve 66,400 cap cases. When the FY 2004 shortfall and Chile/Singapore adjustments were made the 66,400 number dropped to 57,900, the manager reported. Page 18

Also on September 29, 2004, a CIS manager suggested that a draft Federal Register Notice could not possibly be approved in time, and that CIS would need to issue a press release about cutting off further applications on October 1st. According to his e-mail, one of the reasons the Federal Register Notice had not yet been approved was that "issues related to the Chile and Singapore set-asides are still being worked out." It then required two more days to obtain all the required CIS, DHS, and other agency clearances before the press release was final. It was issued on October 1, 2004. 15 Recent CIS initiatives are intended to prevent a recurrence CIS is concerned that it had neither a technology nor a methodology to adhere to the legislated cap. Managers told us that its FY 2005 failure has been the subject of several discussions with officials at the highest level in DHS. It has initiated two new measures to prevent a recurrence: bi-specialization and last-day randomization. Bi-specialization As we described earlier in this report, because H-1B petitions are processed at all four service centers, any effort to collect data requires sweeping four computer systems. Any procedural changes need to be implemented among four workforces, and new policy decisions need to be discussed with four managements. CIS has begun to centralize specific processes at specific service centers to reduce these difficulties. In October 2004, a discussion memorandum was prepared in the CIS Office of Service Center Operations that analyzed the issue and made recommendations for reorganization. The paper acknowledged that previous efforts to "specialize" production had run into resistance due to the concerns of "managers and supervisors that some work is more attractive than other work due to the grade appropriateness and/or the high profile nature of some cases." The proponents of increased specialization have made efforts to avoid any negative effect on employee numbers and grade levels. Among the foreseen benefits of increased specialization are: increased quality and consistency of adjudications, improved fraud detection, enhanced customer service, budgetary savings, and improved production management and resource allocation. In the discussion of production management improvements, the memorandum specifically states that bi-specialization will "facilitate management of congressionally mandated numerical caps". 15 A copy can be found at http://uscis.gov/graphics/publicaffairs/factsheets/h1b_05fnl100104.pdf. Page 19

Bi-specialization will be introduced over a two-year period. Each specific CIS product, such as temporary worker visa petitions, will be processed at only two service centers. As a result, each of the centers will have half as many types of products to produce. When we asked why H-1B processing would not be centralized at just one service center, further improving cap management, CIS managers stated that the existing CIS computer systems are neither robust nor redundant, so it would be unwise to give sole responsibility for any product to any one service center for fear that a computer outage at that center would shut down an entire product line. Last-day Randomization CIS describes this new procedure in a May 4, 2005, press release: For FY 2006 and subsequent fiscal years... USCIS will accept and adjudicate properly filed H-1B petitions on a first in, first out basis using projections that indicate the number of petitions necessary to reach the congressionally mandated cap. USCIS will closely monitor that number and notify the public of the final receipt date (the date USCIS receives the necessary number of petitions to meet the cap). For petitions received on the final receipt date, USCIS will apply a computer-generated random selection process. This process will randomly select the exact number of petitions from the day s receipts needed to meet the congressionally mandated cap. USCIS will reject all petitions received on the final receipt date not selected through the random process. In the interest of fairness and orderly procedures, if any cap is reached on the first day on which filings can be made, the random selection will include the filings received on that first day AND the following day. 16 Last-day randomization may not resolve the counting problem. It has the appearance of transparency and fairness. But nothing we observed of the petition approval and counting process gives us confidence that CIS can know in advance the date on which it is likely to approve its 65,000th cap case. Indeed, for all the reasons described in this report, it is unclear even now on exactly which date CIS approved FY 2005 cap case number 65,000. Conclusions and Recommendations The number of persons granted immigrant status each year is strictly controlled, whether they are issued visas abroad by DOS, or adjust status in the United States 16 Press Release, "USCIS Implements H-1B Visa Reform Act Of 2004; Announces New H-1B Procedures For FY 2005 and FY 2006," dated May 4, 2005. Observed at : http://uscis.gov/graphics/publicaffairs/newsrels/h-1b_050504.pdf. Page 20